Background: Inappropriate prescription is a relevant problem in primary health care settings in Mexico, with potentially harmful consequences for patients. Objective: To evaluate the effectiveness of incorporating a pharmacist into primary care health team to reduce prescription errors for patients with diabetes and/or hypertension. Setting: One Family Medicine Clinic from the Mexican Institute of Social Security in Mexico City. Method: A "pharmacotherapy intervention" provided by pharmacists through a quasi experimental (before-after) design was carried out. Physicians who allowed access to their diabetes and/or hypertensive patients' medical records and prescriptions were included in the study. Prescription errors were classified as "filling", "clinical" or "both". Descriptive analysis, identification of potential drug-drug interactions (pD-DI), and comparison of the proportion of patients with prescriptions with errors detected "before" and "after" intervention were performed. Main outcome measure: Decrease in the proportion of patients who received prescriptions with errors after the intervention. Results: Pharmacists detected at least one type of error in 79 out of 160 patients. Errors were "clinical", "both" and "filling" in 47, 21 and 11 of these patient's prescriptions respectively. Predominant errors were, in the subgroup of patient's prescriptions with "clinical" errors, pD-DI; in the subgroup of "both" errors, lack of information on dosing interval and pD-DI; and in the "filling" subgroup, lack of information on dosing interval. The pD-DI caused 50 % of the errors detected, from which 19 % were of major severity. The impact of the correction of errors post-intervention was observed in 19 % of patients who had erroneous prescriptions before the intervention of the pharmacist (49.3-30.3 %, p < 0.05). Conclusion: The impact of the intervention was relevant from a clinical point of view for the public health services in Mexico. The implementation of early warning systems of the most widely prescribed drugs is an alternative for reducing prescription errors and consequently the risks they may cause.